Repeatability limits for measurement of coronary artery calcified plaque with cardiac CT in the multi-ethnic study of atherosclerosis

OBJECTIVE. The purposes of this study were to examine the repeatability of the findings of coronary artery calcification (CAC) measured with CT on repeated scans, to estimate 95% repeatability limits for CAC, and to use these limits to quantify detectable change in CAC over time. SUBJECTS AND METHODS. The Multi- Ethnic Study of Atherosclerosis is a prospective cohort study with 6,814 participants 45-84 years old and free of clinical cardiovascular disease at enrollment. Agreement for presence of CAC was assessed for 6,742 participants who had baseline replicate scans on which a CAC score of 0 indicated no coronary calcification. Among 3,380 participants with baseline CAC, the 95% repeatability limits were established with a quantile regression model. Detectable change in CAC during follow-up was defined by an increase or decrease beyond the baseline repeatability limit. RESULTS. At baseline, 274 (4.1%) of the rescan pairs were discordant (presence or absence of CAC). Greater body mass index was associated with a discordant pair (trend, p < 0.05). The upper 95% repeatability limits were (0.17 x Agatston score) + (4.89 x root Agatston score) + (0.44 x body mass index) - 10.84 for Agatston score and (0.16 x volumetric calcium score) + (4.30 x root volumetric calcium score) + (0.23 x body mass index) - 5.00 for volumetric calcium score. Rescan repeatability was comparable for electron beam and 4-MDCT scanners. At 2.5 years of average follow-up (range, 0.9-5.0 years), a detectable increase in Agatston and volumetric calcium scores was observed in 1,027 (36.3%) and 1,020 (36.0%), respectively, of 2,832 participants with baseline CAC. CONCLUSION. The repeatability limits derived can be used to evaluate whether an increase in CAC score exceeds that expected from measurement error alone.